Deep learning tracks Paris urban change over decades
Summary and headline written by AI from the source article. How we work
Researchers developed a deep-learning framework that automatically analyzes historical maps to reveal patterns of urban change.
The system aligns maps, extracts objects like buildings and roads, and then estimates changes over time. This new approach moves beyond simple visual comparisons of old maps, allowing for detailed, quantitative study of how cities evolve. The framework successfully processed a series of historical maps of Paris spanning 1868 to 1937, uncovering variations in the speed and location of urban transformation.
The modular design means the system could be adapted to analyze maps from other cities and for different research questions. The team suggests this method could be valuable to social scientists and those in the humanities, providing data for understanding urban history. A key challenge remains in dealing with the imperfections of old maps, spatial distortions and fading quality, but the system demonstrates a robust ability to overcome these issues.
The researchers envision this technology providing a new way to study long-term urban development and its impact on societies.
